<p>The Metaverse feels like a reflection of reality, offering immersive experiences where people can interact, explore, and connect as they do in the physical world. It appears in a virtual environment that mimics the actual world. Metaverse technology improves energy efficiency, material, and human resource utilization, and sustainable development. Avatars are people’s representations in the Metaverse. The world of the Metaverse recently received a lot of attention from the fashion industry. Fitting garments virtually on an avatar in Metaverse fashion has been a transformative technology that supports users to choose the best outfit suitable for the avatar and occasion in a virtual environment. In this paper, a model-based adaptive StyleGAN is proposed to dress up the human avatar with various garments. It also supports users in trying on different garments through virtual avatars without needing to wear them physically. The proposed model starts by creating the avatar with different fashion styles using the StyleGAN network. Then, the garments are extracted using the K-means clustering algorithm. It’s followed by fitting the clothes on the human avatar. StyleGAN2 is a kind of StyleGAN that is used to generate humans with different fashion styles. These StyleGANs are StyleSpace, InterFaceGAN, and SeFa are a type of StyleGAN2, and they are proposed to generate human avatars with different fashion styles. To evaluate the generated images from the three StyleGANs, Blind-Referenceless Image Spatial Quality Evaluator (BRISQUE), Naturalness Image Quality Evaluator (NIQE), and Perception-based Image Quality Evaluator (PIQE) are applied. The evaluation showed that, first, the results are unaffected by the quantity of tested images. Additionally, the BRISQUE gives the best result for SeFa GAN, the StyleSpace is the optimal result for NIQE, and InterFaceGAN is the optimal result for PIQE. The results are evaluated using the Mann-Whitney U test.</p>